delivered the opinion of the court:
Defendant Diane Kent was charged by indictment with the murder of her four-month-old daughter. Count I charged that defendant, in violation of section 9 — 1(a)(1) of the Criminal Code of 1961 (Ill. Rev. Stat. 1979, ch. 38, par. 9—1(a)(1)), knowingly and intentionally killed the baby by feeding the child alcoholic beverages. Pursuant to defendant’s motion at the close of the State’s case, the trial court entered a finding in favor of defendant on this count. Count II charged that *734defendant violated section 9 — 1(a)(2) of the Criminal Code in that she knew her actions created a strong probability of death or. great bodily harm. Defendant was found guilty on this count and sentenced to 20 years in the Illinois Department of Corrections.
On appeal, defendant contends (1) she was not proved guilty beyond a reasonable doubt; (2) the trial court erred in assessing the qualifications of a defense expert on evidence outside the record'; (3) testimony regarding the results of tests performed on a tissue specimen was improperly admitted; and (4) the trial court erred in denying defendant’s motion for substitution of judges.
We reverse.
At trial, Officer McLaughlin testified that at about 12:30 p.m. on June 23, 1980, he arrived at defendant’s home. Paramedics at the scene stated the infant had no vital signs. Officer McLaughlin and his partner took the baby to Provident Hospital where at 1:15 p.m. the baby was pronounced dead. The deceased was then taken to Cook County Morgue.
Officer Michael Pochordo, a Chicago Police Department detective conducted a follow-up investigation on the baby’s death. On July 26, 1980, Pochordo spoke to Dr. Donoghue, the county pathologist, who told him the baby died of acute alcoholism. He interviewed defendant and her sister the following day at the police station. During the interview, defendant stated she was an alcoholic. Pochordo immediately called assistant State’s Attorney Schwind who interviewed defendant in Pochordo’s presence. On cross-examination, Pochordo testified that Dr. Donoghue initially attributed the cause of death to Sudden Infant Death Syndrome.
Richard Schwind testified that on July 27, 1980, Pochordo informed him of the facts of the baby’s death. He spoke to Dr. Donoghue and defendant’s sister and then interviewed defendant. During the interview, defendant agreed to give a written statement. This statement, given at 1:15 p.m. on July 27, 1980, was read into evidence. The following facts are found in the statement.
Defendant lived with her grandmother, aunt, sister and her sister’s three children. On June 22, 1980, defendant arose at 7 a.m. She fed the baby creamed corn and a bottle of milk. She also gave the baby liquid Tylenol because the baby had a fever. Defendant went to church and returned home at 2:30 p.m. She left again that afternoon, returned home at 9:30, fed the baby milk, and took her for a walk. The baby was sleeping when they arrived home. Thereafter, defendant and her sister’s boyfriend went to a bar where defendant drank four beers. At about midnight, defendant’s brother came to the bar to *735bring defendant home because the baby was crying. Defendant fed the baby a bottle of milk prepared by her sister. She and the baby fell asleep. At about 2 a.m. on June 23, 1980, the baby was crying, but would not drink the milk defendant tried to feed her. When defendant arose at noon, she lifted the baby. The baby was cold to the touch. Defendant also stated she was an alcoholic and attended Alcoholics Anonymous meetings. Twice daily from the time the baby was three weeks old, defendant gave the baby beer mixed with equal parts of water. She could not recall if she fed the baby beer on June 22 or 23.
Dr. Donoghue, physician, forensic pathologist, and deputy chief medical examiner of Cook County, performed an autopsy on Mary Jane Kent on June 24, 1980. He testified that the deceased weighed 12 pounds and measured 221k inches in length. These measurements indicated the baby was undernourished and suffered growth retardation. Ninety-seven percent of babies the age of the deceased weighed more and were taller than the deceased. Upon performing an internal examination, Dr. Donoghue found petechial hemorrhages on the lung surfaces and pulmonary edema fluid in the trachea. The liver was enlarged and pale tan in color. The stomach contained tan fluid. A microscopic examination of internal organs revealed fat vacuoles in the liver and aspiration of gastric content in the lungs. Aspiration occurred when stomach fluid traveled through the esophagus and trachea into the lungs where the fluid caused chemical damage. Aspiration prevents lungs from absorbing oxygen, thereby causing asphyxiation.
Dr. Donoghue also requested an alcohol determination on muscle tissue he removed from the neck of the deceased. The test is customarily performed on blood samples. However, he initially did not consider alcohol poisoning as a possible cause of death and thus preserved no blood from the body. The muscle tissue was sent to the toxocology lab. Test results revealed 20 milligrams percent ethanol in the tissue. Dr. Donoghue determined death was caused by aspiration of gastric content due to alcohol intoxication. Significant contributing factors were fatty liver and chronic alcoholism.
On cross-examination, Dr. Donoghue testified that the time of death was not ascertained exactly, but death occurred between 3 a.m. and 12:45 p.m. on June 23, 1980. He further testified that decomposition after death produces ethanol. If ethanol is produced post-mortem, it usually is accompanied by a bad odor. No similar odor is present if ethanol is produced from consumption of alcohol. The amount of ethanol found in body organs and fluid varies if ethanol production occurs post-mortem, whereás if ethanol is found in equal levels in different *736organs, its presence is not caused by decomposition.
Donoghue further testified on cross-examination that he took only one muscle sample and although he took the sample from the neck, several other muscles would have yielded suitable samples. When alcohol is consumed, its concentration first rises in the blood, then in the muscle, and is eventually equalized in all parts of the body. Generally, alcohol is eliminated from the system in 24 hours. The average ethanol level in those who die from alcohol poisoning is about 350 milligrams percent. Dr. Donoghue also stated a fatty liver is a manifestation of malnutrition and Reyes Syndrome. Other significant pathological findings in this case are also commonly found in children who succumb to Sudden Infant Death Syndrome.
Dr. Michael Schaffer, chief toxocologist for the office of the medical examiner of Cook County, supervised the laboratory at the morgue. He testified to the chain of custody of the muscle sample and described the procedures used to test specimens for volatiles. The muscle sample in this case was received by him at 2:49 p.m. on June 23, 1980. The tissue was not frozen. The tests were performed about 24 hours later. The method used to detect alcohol content does not distinguish the cause for presence of ethanol. The tissue sample was buried some time after the tests were performed.
Sofia Dziugas, a laboratory technician in the medical examiner’s office, performed tests on the muscle sample and described the testing procedures. The test was run three times to verify the results. The results were 19.5, 19.6 and 19.8 milligrams percent ethanol.
Dr. Leo Goldbaum, a toxocology expert called by the defense, testified that it is common to submit more than one tissue sample for examination in order to eliminate inaccurate findings due to specimen contamination. Alcohol contamination of specimens may occur when the specimen comes into contact with laboratory equipment from which disinfectant has not been adequately removed. The autopsy table and specimen jars are also sources of contamination. Almost every specimen from a post-mortem examination is contaminated by one or more microorganisms which produce ethanol upon reacting with certain body chemicals. Post-mortem contamination or putrefaction begins immediately, occurs rapidly, and cannot be detected by sight and smell alone. Refrigeration does not totally retard the process of decomposition. Determining whether ethanol is produced ante-mortem or post-mortem is difficult. However, if some specimens have a higher concentration than others, post-mortem contamination is indicated.
Dr. Goldbaum further testified that in cases where death is caused by acute alcohol intoxication, the concentration is about 400 milli*737grams percent. In adults, 20 milligrams percent alcohol would be an insignificant finding and would not have affected the physical state of the person. The important factor in alcohol’s effect on the central nervous system is the concentration in the body, not the amount consumed or the person’s height and weight. Dr. Goldbaum knew of no studies regarding the effects of alcohol on young people. However, experiments performed on small animals, some of which weighed about 12 pounds, have shown that 20 milligrams percent alcohol had no adverse effects. Further testimony revealed that prescription and nonprescription elixirs, including some medications designed for children, contain alcohol.
Defendant also called Dr. David Petty as an expert witness. At the time of trial, Dr. Petty was a physician and surgeon. He had received a degree in pathology and had been employed as a deputy coroner physician in Cook County for 10 years, during which time he performed over 500 autopsies. Several of Dr. Petty’s articles on cirrhosis of the liver had been published.
Dr. Petty testified that young children have a particularly low resistance to the virus called Reyes Syndrome. Pathological findings in a child who had died from this disease would be a fatty liver, congestion and edema in the lungs, and conjunction and edema of the brain. Malnutrition, alcoholism and various toxic conditions also cause fat vacuoles to develop in the liver. To help determine the cause of fatty liver, a Mallory test is often performed. In this test, liver tissue is stained with a special solution and microscopically examined for the presence of Mallory hyalines (hyaline membranes in liver cytoplasm). Hyalines are present in cirrhotic livers. However, they are not present in fatty livers associated with Reyes or other diseases.
Dr. Petty further testified that the generally accepted practice in testing muscle samples for the presence of ethanol or alcohol is to remove a sample from each of three different muscles. If test results on each sample are identical, the presence of alcohol is caused by alcohol ingestion. Considerable variance in test results would indicate production of ethanol from putrefaction. When death is caused by acute alcohol intoxication, the concentration of ethanol is about 300 to 350 milligrams percent. A low level of alcohol such as that found in the deceased in this case could be attributed to putrefaction.
According to Dr. Petty, blood specimens are preferable to other tissue samples and, if blood is not available, tests are often performed using the vitreous humor. Other tissue samples may be contaminated by bacteria from the skin surface, autopsy instruments, the autopsy table or the hands of the pathologist, but the vitreous humor gener*738ally escapes contamination. Dr. Petty knew the medical examiner’s office performed tests on vitreous humor and had been doing so for several years.
Dr. Petty also testified that several elixirs are prescribed for young childrens’ illnesses. Children’s Tylenol is one of these. The volume of alcohol in elixirs varies from 7 to 11 percent. The alcohol content of beer is between 3.2 and 6 or 7 percent. In response to a hypothetical question assuming the facts presented in the case,- Dr. Petty stated there was insufficient evidence to conclude that the deceased died from acute alcohol intoxication. Dr. Petty explained the low level of ethanol in the muscle tissue was within the range found in a person who never consumed alcohol and did not indicate alcohol intoxication. Only eight to ten percent of those who consume large quantities of alcohol develop cirrhotic livers. Cirrhosis takes some time to develop. The fatty liver of the deceased could have been caused by various diseases, and these causes were not excluded by proper testing.
First, defendant contends she was not proved guilty beyond a reasonable doubt. She maintains the evidence adduced at trial did not prove beyond a reasonable degree of medical and scientific certainty that her actions of feeding beer to the baby caused the baby’s death.
In every murder prosecution, proof of death and proof of criminal agency causing death are elements the State must prove beyond a reasonable doubt. (People v. Guthrie (1980), 85 Ill. App. 3d 831, 407 N.E.2d 593; People v. Brown (1978), 57 Ill. App. 3d 528, 373 N.E.2d 459.) Although it is not necessary to prove that defendant’s act is the sole and immediate cause of death (Brown), the State must nonetheless prove that defendant’s act was beyond a reasonable doubt a contributing cause such that death did not result from a cause unconnected with defendant. (Brown, citing People v. Brown (1973), 9 Ill. App. 3d 730, 293 N.E.2d 1.) The means and manner may be inferred from the circumstances proved, but guilt must be so established as to exclude reasonable hypotheses of innocence. People v. Kinzell (1969), 106 Ill. App. 2d 349, 245 N.E.2d 319.
In the case at bar, the evidence was insufficient to prove beyond a reasonable doubt that defendant’s acts were the cause of death. First, all expert witnesses, including the State’s expert witness Dr. Petty, agreed that Sudden Infant Death Syndrome, Reyes Syndrome, and malnutrition would cause the physical conditions found during the autopsy. The fact the deceased was ill with a fever the day before death only supports the theory that death was caused by another aliment. The Mallory test could have conclusively established whether alcoholism, rather than one of the aforementioned diseases, *739was a significant contributing cause of death, yet this test was not performed. Second, the low level of ethanol in the muscle specimen was not proved to be the result of alcohol consumption. Substantial evidence was presented indicating that ethanol could have been produced by decomposition. The time of death was not ascertained and may have occurred almost 12 hours before the deceased was taken to the morgue. Furthermore, the liquid Tylenol given to the deceased the day before her death had a greater alcohol content than most domestic beers. This substance or possible bacterial contamination of the specimen could have been the cause of the presence of ethanol in the muscle tissue. Third, the only evidence regarding the quantity of alcohol given the deceased over a period of time was defendant’s statement to Schwind that she fed the baby beer mixed with equal parts of water. The amount consumed and its effect were never proved. According to the testimony of Dr. Goldbaum regarding experiments performed on small animals, small quantities of alcohol would have little or no adverse effects. Fourth, defense experts testified that commonly accepted pathological procedures dictate that three muscle samples should be examined to determine the cause of ethanol production and eliminate the possibility of specimen contamination. Failure to follow this procedure in this case leads to the conclusion that the determination regarding the cause of death was mere conjecture.
We recognize that reasonable hypotheses compatible with innocence need not be elevated to the status of reasonable doubt. (People v. Tackett (1980), 91 Ill. App. 3d 410, 414 N.E.2d 748.) However, in this case the evidence that death was caused by a criminal agency is so unsatisfactory that we are compelled to reverse the finding of guilt.
Defendant also contends she was denied a fair trial where the trial court considered matters outside the record in determining the credibility of Dr. Petty. Defendant offered Dr. Petty as both a physician and pathologist. The trial court conducted an in camera proceeding to determine Petty’s qualifications as a pathologist. During the hearing the trial court made the following comments:
“THE COURT: Counsel, those of us that have' been associated with the courts in this country know, and this is common knowledge and I think we all know, that under the time when he worked for the coroner’s office, the office was not run very efficiently, there were many people connected with that office that were not qualified in my opinion. Now, if he is —
MR. McGANN: — Judge, I am going to right now, I will move for a mistrial.
*740THE COURT: Why? This is common knowledge.
* * *
THE COURT: — you brought in this man. I know what I know about the coroner’s office. Let me say this: If I were the State’s Attorney I would ask him some more questions. In those days there were doctors that used to come in there on a per diem basis that would be practicing and then would just do
one body and go off on their practice.
* * *
THE COURT: Let’s put it this way, my observations of the office was such that at that time he was not, it was not a good office. It was a very politically run office. You have to overcome that little bit of knowledge that I have about the office.”
The trier of fact is limited to considering only evidence presented at trial and cannot base its determinations on private knowledge. (People v. Wallenberg (1962), 24 Ill. 2d 350, 181 N.E.2d 143.) In a bench trial, it is presumed the trial court considered only competent evidence, but the presumption may be rebutted by affirmative evidence in the record. (People v. Gilbert (1977), 68 Ill. 2d 252, 369 N.E.2d 849.) The statements of the trial judge in this case clearly indicate that personal knowledge was considered in assessing the credentials and credibility of Dr. Petty. We are left with no doubt that the credibility of Dr. Petty was minimized. Additionally, defendant was burdened with the unfair and perhaps impossible task of overcoming the personal knowledge of the court, which should not have been considered. In this case, defendant was denied a fair trial.
Our decision in this case renders consideration of the other issues raised by defendant unnecessary.
For the foregoing reasons, the judgment of the circuit court of Cook County is reversed.
CAMPBELL, P.J., and GOLDBERG, J., concur.